NixOS and Computational Reproducibility: Why you won’t need to install software anymore soon.
IPGP - Îlot Cuvier
Séminaires de Sismologie
Computational reproducibility is a big topic in science. A reproducibility crisis has even been proclaimed backed by a multitude of studies showing that a large part of published computational results can't be reproduced, and that's true even if data and source code are available! And it's unfair to blame the scientists for this: A regular 10-year reproducibility challenge shows how difficult it actually is to write software in a way that someone else can execute it and get the same result. Established researchers know how time consuming it is for new PhDs to simply run the exact same code that their predecessors have left behind. I'll explain a bit why it is so surprisingly difficult to make a machine that we have full control over do exactly what we want, and I'll go into modern solutions to the reproducibility problem, explaining and demoing the next-gen Linux NixOS. Participer à la réunion Zoom https://cnrs.zoom.us/j/92698340794?pwd=RHFlRm13c1JtSTJFQ2I5Z3lsdFk1QT09 ID de réunion : 926 9834 0794 Code secret : zA7Wv8